The C-mean clustering algorithm is modified and adapted in order to be applicable in the current study. In addition, six groups are defined for different focal mechanisms and soil types. Then, the GMPE clustering is performed for each group and the obtained clusters are proposed and discussed. The results confirm that the obtained spectral ordinated from GMPEs of different clusters can meaningfully differed from each other. | JSEE Vol. 19, No. 1, 2017 Adaptive Fuzzy C-Mean Clustering of Ground Motion Prediction Equations Alireza Azarbakht1* and Zahra Minaei 2 1. Associate Professor, Department of Civil Engineering, Faculty of Engineering, Arak University, Iran *Corresponding Author; email: a-azarbakht@ 2. . Student, Department of Civil Engineering, Faculty of Engineering, Arak University, Iran Received: 17/02/2016 Accepted: 16/05/2017 ABSTRACT Keywords: Earthquake; GMPE; Fuzzy C-mean clustering; Un-supervised classification; Validity index; Next Generation Attenuation (NGA) Selection of Ground Motion Prediction Equations (GMPEs) within the Seismic Hazard Analysis (SHA) is an important and timely research line of inquiry. A set of 22 regional and worldwide GMPEs have been selected in this research for the purpose of classification. They are classified into clusters in which each cluster is defined to have the most dissimilarity with the other clusters as well as having the most similarity within the cluster. The C-mean clustering algorithm is modified and adapted in order to be applicable in the current study. In addition, six groups are defined for different focal mechanisms and soil types. Then, the GMPE clustering is performed for each group and the obtained clusters are proposed and discussed. The results confirm that the obtained spectral ordinated from GMPEs of different clusters can meaningfully differed from each other. 1. Introduction Ground Motion Prediction Equation (GMPE) is the key element within any seismic hazard analysis (SHA). Different GMPEs have been revised by enrichment of earthquake catalogues in which a wide variety of GMPEs are now available [1-2]. However, any GMPE depends strongly on the selected ground motion database. In other words, employing non-local GMPEs may be a challenging task [3]. Additionally, selection of a suitable GMPE is usually performed without enough attention to the compatibility of the chosen GMPE on historical site .
